[tests] make `test_trainer_log_level_replica` to run on accelerators with more...

[tests] make `test_trainer_log_level_replica` to run on accelerators with more than 2 devices (#29609)

add new arg

tests/extended/test_trainer_ext.py

@@ -62,6 +62,7 @@ class TestTrainerExt(TestCasePlus):
         do_train=True,
         do_eval=True,
         do_predict=True,
+        n_gpus_to_use=None,
     ):
         output_dir = self.run_trainer(
             eval_steps=1,
@@ -74,6 +75,7 @@ class TestTrainerExt(TestCasePlus):
             do_train=do_train,
             do_eval=do_eval,
             do_predict=do_predict,
+            n_gpus_to_use=n_gpus_to_use,
         )
         logs = TrainerState.load_from_json(os.path.join(output_dir, "trainer_state.json")).log_history
 
@@ -138,7 +140,13 @@ class TestTrainerExt(TestCasePlus):
         }
 
         data = experiments[experiment_id]
-        kwargs = {"distributed": True, "predict_with_generate": False, "do_eval": False, "do_predict": False}
+        kwargs = {
+            "distributed": True,
+            "predict_with_generate": False,
+            "do_eval": False,
+            "do_predict": False,
+            "n_gpus_to_use": 2,
+        }
         log_info_string = "Running training"
         with CaptureStderr() as cl:
             self.run_seq2seq_quick(**kwargs, extra_args_str=data["extra_args_str"])